1、多属性决策分析案例(分析过程文件见文件夹“分析过程”)第一步：利用MATLAB求综合属性值M-FILE 源文件见 “multiattribute.m”代码：%first import the data for A!%首先输入数组 A（即文件“多属性决策”）! 源数据顺序经过了调整，把逆指标放在前23列，正指标放在后八列R=zeros(60,31);R1=zeros(60,31);E=zeros(1,31);W=zeros(1,31);Z=zeros(60,1);B=ones(1,31);%把B 变成全为1的数组k=1:23;B(1,k)=1./max(A(:,k)-min(A(:,k); %确
2、定MAX(a(i,j)-MIN(a(i,j)数组for j=1:23R(:,j)=(max(A(:,j)-A(:,j).*B(1,j);%先把前23个逆指标规范化 endB=ones(1,31);%再次把B 变成全为1的数组k=24:31;B(1,k)=1./max(A(:,k)-min(A(:,k); for j=24:31R(:,j)=(A(:,j)-min(A(:,j).*B(1,j)%再把后八个正指标规范化for j=1:31R1(:,j)=R(:,j)./(sum(R(:,j);%将矩阵R列归一化ende=0;for j=1:1:31for i=1:1:60if R1(i,j)=0s
3、=0;else s=R1(i,j)*log(R1(i,j);ende=s+e;endE(j)=(-1/log(60)*e;%得出信息熵endsK=sum(1-E(1,:);for j=1:31W(j)=(1-E(j)./sK;%得出权重endfor i=1:60;z=0;for j=1:31z=z+R(i,j)*W(j);endZ(i)=z;%得出综合属性值Zendend综合属性值 Z的具体数值见文件 EXCLE “综合属性 Z的数值 “第二步：用SPSS进行聚类分析分析结果聚类成员案例号 VAR00002 聚类 距离1 1.00 1 .0132 2.00 5 .0023 3.00 4 .01
4、44 4.00 4 .0015 5.00 4 .0316 6.00 4 .0187 7.00 5 .0248 8.00 5 .0179 9.00 5 .02210 10.00 4 .01211 11.00 2 .01812 12.00 3 .00613 13.00 5 .01614 14.00 4 .00015 15.00 4 .01416 16.00 5 7.511E-617 17.00 2 .00118 18.00 2 .01019 19.00 2 .02220 20.00 1 .01921 21.00 4 .03722 22.00 3 .00223 23.00 1 .00624 24.0
5、0 4 .03625 25.00 2 .00226 26.00 4 .01627 27.00 4 .01928 28.00 3 .00129 29.00 4 .01030 30.00 2 .03031 31.00 2 .01432 32.00 5 .00233 33.00 3 .00934 34.00 4 .02435 35.00 4 .02336 36.00 2 .01737 37.00 5 .01038 38.00 4 .00239 39.00 2 .01640 40.00 5 .00741 41.00 2 .00242 42.00 4 .00543 43.00 2 .00744 44.0
6、0 4 .00345 45.00 2 .03246 46.00 4 .02147 47.00 4 .02448 48.00 2 .02349 49.00 2 .00650 50.00 2 .01251 51.00 2 .01752 52.00 2 .01753 53.00 5 .02454 54.00 5 .00755 55.00 1 .01256 56.00 5 .02057 57.00 5 .00758 58.00 5 .01359 59.00 4 .02160 60.00 2 .018最终聚类中心聚类1 2 3 4 5综合属性值Z .49 .71 .56 .65 .76每个聚类中的案例数1 4.0002 18.0003 4.0004 20.000聚类5 14.000有效 60.000缺失 .000
